docs: index the jar locator and stop promising exit 1 means CI #514

Merged
mph merged 1 commit from docs/skills-0.13.0 into main 2026-09-03 09:43:28 +00:00
Owner

phase-pr gained a refusal in #511 that exits 1 having gated nothing — the branch already carries a merged or closed pull request, or the forge would not say which request is on it. The published exit contract said exit 1 means "CI refused the phase", so an agent hitting the new refusal would hunt for a CI failure and a job log that do not exist.

The contract now names the three situations exit 1 covers, and says which one leaves no log. It also states that the phase branch was pushed before that refusal: a reader who takes "the command could not run" to mean nothing happened is wrong about that, and the consumer confirmed they would have documented it incorrectly without being told.

The same false line exists in the downstream plugin's own pipeline blueprint, which generates into six skills; that is tracked on their side and gated on this release.

Also: llms.txt and the command-creation module table gained CoreJar and InstalledJar, both public and both new in #511, with a docs/api/CoreJar.md page. Process.md no longer describes a stderr line that runStreaming stopped writing.

🤖 Generated with Claude Code

https://claude.ai/code/session_013UpRzj7oiV1UDArn4hfu7U

`phase-pr` gained a refusal in #511 that exits 1 having gated nothing — the branch already carries a merged or closed pull request, or the forge would not say which request is on it. The published exit contract said exit 1 means "CI refused the phase", so an agent hitting the new refusal would hunt for a CI failure and a job log that do not exist. The contract now names the three situations exit 1 covers, and says which one leaves no log. It also states that the phase branch *was* pushed before that refusal: a reader who takes "the command could not run" to mean nothing happened is wrong about that, and the consumer confirmed they would have documented it incorrectly without being told. The same false line exists in the downstream plugin's own pipeline blueprint, which generates into six skills; that is tracked on their side and gated on this release. Also: `llms.txt` and the command-creation module table gained `CoreJar` and `InstalledJar`, both public and both new in #511, with a `docs/api/CoreJar.md` page. `Process.md` no longer describes a stderr line that `runStreaming` stopped writing. 🤖 Generated with [Claude Code](https://claude.com/claude-code) https://claude.ai/code/session_013UpRzj7oiV1UDArn4hfu7U
docs: index the jar locator and stop promising exit 1 means CI
All checks were successful
CI / contract (pull_request) Has been skipped
CI / format (pull_request) Successful in 7s
CI / lint (pull_request) Successful in 7s
CI / compile (pull_request) Successful in 1m16s
CI / dashboard-build (pull_request) Successful in 1m1s
CI / test (pull_request) Successful in 9m32s
1dd8a1d009
`phase-pr` gained a refusal that exits 1 having gated nothing: the branch
already carries a resolved pull request, or the forge would not say which
request is on it. The published exit contract said exit 1 means CI refused
the phase, so an agent hitting the new refusal would look for a CI failure
and a job log that do not exist.

The contract now names the three situations exit 1 covers and says which
one leaves no log — including that the phase branch was pushed, since a
reader who takes "the command could not run" to mean nothing happened is
wrong about that.

`llms.txt` and the module table gained CoreJar and InstalledJar, and
`Process.md` no longer describes a stderr line `runStreaming` stopped
writing.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013UpRzj7oiV1UDArn4hfu7U
mph merged commit 6dcdd965fc into main 2026-09-03 09:43:28 +00:00
Sign in to join this conversation.
No description provided.